Commentary
The Accountability Gap: Can Offline Sales Actually Be Linked To Digital Ads?
- by Abraham Snyder , March 4, 2014
Not to be the voice of pessimism, but I’d like to take this opportunity to emphatically remind marketers that despite what vendors say, there are no magic bullets in the marketplace that would firmly link ads to offline sales! Not to be a total killjoy, I can say that there has been enough innovation in the marketplace, where we are definitely getting better at offering brands a nuanced, albeit directional indication that links specific campaigns to offline sales. It is a rare scenario, however, when all the data pieces come together where one can attribute specific impressions or engagements to specific sales.
So it is incumbent upon my agency brethren to take firmer stances with their clients and offer them a reality check.
What Is Actually Possible?
What is possible is all over the place. Loyalty cards and programs have been a significant contributor to data collection for retailers for a long time. For brands, however, what’s available is disparate and often not timely enough to be actionable. Unfortunately, a major roadblock to being able to definitively connect media spend to offline sales can be found in the resistance on the part of the big guys -- Target and Walmart -- to share timely and actionable data with the very brands selling on their shelves. These two retailers represent nearly 23% of the retail market (KantarRetail 2013), and their dominance is even more amplified if you narrow the focus to specific categories like CPG.
But could we maybe look away from the retailers and to the consumers themselves to reconcile on a household level? For example, Nielsen Homescan represents about 100,000 households in the U.S., and is used for projecting out against the nation as a whole. With digital, people expect a one-to-one of exposure-to- purchase; we’re so far from that with the Homescan projections. All of this leaves us with a partial view of the marketplace, and even that is a view with a lot of holes. This incomplete picture makes performing purchase analysis on an individual level not possible. We will always be able to perform market testing and other analysis techniques to understand general lift of a campaign on sales, but moving beyond this norm is still proving to be a challenge.
It is important that agencies take a brave position around this issue with their clients. It’s incumbent on agencies to set realistic expectations now or suffer the consequences of unmet ones down the road.
It’s not all bad, though. Some companies are setting up the backbone that will at some point make this all possible. The work they are doing to create plug-and-play data platforms that make the tough correlations between disparate sources of data is invaluable, and will surely spawn other packaged solutions. It’s possible for some brands in some industries to use these platforms and get nearly real-time, actionable insights. However, as long as the availability of data is limited, the dream will remain elusive to most, and the accountability gap will persist.
Check out Punchcard. Shopping app enables consumers to take pictures of receipts and be marketed to based on their item level preferences, everywhere they shop. Great accountability when combined with a media play.
We have a robust measurement model that causally measure TV commercial ad effects and offline CPG purchases for over 15 years.
We have conducted over 5,300 experimental studies for over 100 major CPG brands.
See: www.EffectiveMarketingMgt.com
Rer: "there are no magic bullets in the marketplace that would firmly link ads to offline sales". A "hold-out test" gets very close to this - compare sales when running the ads vs sales when not running the ads, in two equivalent regions or time periods..
This is somewhat broad and complex topic with the answers depending on which media, which verticals, and the types of offline conversions you are wishing to account for. The solutions vary from econometric models, to coupon redemption and loyalty programs, to hold out tests, to modeled user-level attribution techniques. The best approach and the value of the results greatly depends on the situation.
Working with an agency for a big-box retailer, we are able to attribute 12-14% of POS sales directly to digital media (display, search, email, social). While not perfect, the insights are highly informative and are driving new strategies and growth.
Loyalty programs are a very good way to create the link -- the customer gives permission to be tracked and receives value for self-reporting, esp. in the case of CPG. Punchcard is directionally on target for members in terms of a mobile experience; the value proposition is less clear for both members and brands--and it doesn't link ads to purchases which was the point of the article.
A/B testing is the way to get a clean sales read
Yes, testing is critical. But when it is offline advertising, there needs to be a whole set of controls when running in-market experiments because each 1 of 8 factors can undermine the test results if they have been properly controlled for. That's because these 8 factors work together with chain multiplication logic, A function zero at any one factor reduces the entire end result to zero no matter how high all the other factors. The copy, the media reach, the program schedule and continuity all have a critical impact as well as 3 other non-advertising factors. One of these other factors is concurrent price promotions which cancel out ad effects completely.